Context Effects when Reading Negative and Affirmative Sentences - eScholarship
Barbara Kaup, Jana Lüdtke · Proceedings of the Annual Meeting of the Cognitive Science Society · 2006
Context Effects when Reading Negative and Affirmative Sentences Jana Ludtke ([email protected]) Institut fur Psychologie und Arbeitswissenschaft TU Berlin, Franklinstr. 28/29 (FR 2-6) 10587 Berlin Barbara Kaup ([email protected]) Institut fur Psychologie und Arbeitswissenschaft TU Berlin, Franklinstr. 28/29 (FR 2-6) 10587 Berlin Abstract In two experiments, participants were presented with narratives in which a target entity (e.g., water) was introduced. The final sentences affirmed or negated that a particular property held for the target entity (e.g., The water was (not) warm). The linguistic context was manipulated. In Experiment 1, reading times for the negative target sentences (but not those for the affirmative ones) were affected by the context manipulation, with shorter reading times in conditions in which the negated proposition had been explicitly mentioned in the prior text. Experiment 2 demonstrated a similar facilitation effect with respect to inferred propositions. The results of the two experiments support pragmatic considerations concerning the context dependency of the processing difficulty associated with negative sentences. Keywords: Negation; Pragmatics; Context effects Introduction It is well known that negative sentences are relatively hard to process. In various studies with different methods, negative sentences required more processing time and led to higher error rates than the corresponding affirmative sentences. A number of different explanations were discussed in the literature, including explanations based on sentence length, as well as explanations based on grammatical or connotational aspects of negation (for an overview see Kaup, Zwaan, Ludtke, in press). The most promising explanation rests on considerations concerning pragmatic aspects of negation: The contexts in which negative utterances occur are rather limited. Typically, negative statements are uttered when the negated proposition was either explicitly mentioned before by one of the discourse partners or at least constitutes a plausible assumption in the respective context. Thus, negation is used to communicate deviations from what was previously assumed or expected (e.g., Givon, 1978). This leads to the prediction that negative sentences are difficult to process mainly when presented outside of an adequate context. In this case the respective assumption or expectation is not available prior to encountering the negation, and accordingly must be created retroactively. In contrast, when presented in a context in which the negated proposition is available prior to encountering the negation this additional processing step is not required. Accordingly, in this case negative sentences should not take longer to process than their affirmative counterparts (or at least not as much). This prediction was tested in a study by Wason (1965) with respect to non-linguistic contexts. Participants were presented with pictures of eight circles, seven in one color and one in another color. They were then asked to complete affirmative or negative sentence fragments. Negative fragments (e.g., Circle No 3 is not …) took longer to complete than affirmative fragments (e.g., Circle No 3 is....), but this difference was significantly smaller when the negative sentences referred to the circle with the exceptional color than when they referred to one of the seven other circles. Thus, participants profited from an adequate context when processing negative sentences. In a study by Glenberg, Robertson, Jansen & Johnson-Glenberg, (1999) the same prediction was tested, but by manipulating the linguistic context. To our knowledge this is the only study that experimentally investigated the effect of the linguistic context on the processing difficulty associated with negative sentences (for studies investigating other types of context effects with negation, see Giora, 2006). In Glenberg et al.’s study, participants were presented with longer narratives in which the penultimate sentence was either negative or affirmative (e.g., The couch was / was not black). In the supportive-context condition the previous text highlighted the relevancy of the attribute dimension that was referred to in the penultimate sentence (e.g., She wasn’t sure if a darkly colored couch would look the best or a lighter color). In contrast, in the non-supportive context condition, the prior text highlighted a different attribute dimension (e.g., She wasn’t sure what kind of material she wanted the couch to be made of). Glenberg et al. analyzed the reading times of the penultimate sentence, after correcting for sentence length. In line with the predictions, negative sentences took longer to process than affirmative ones in the non- supportive contexts only. The goal of the present work was to further investigate the question of what constitutes an adequate context for the processing of negative sentences. We focused on the idea that negative sentences are particularly felicitous in contexts in which the proposition that is being negated was either explicitly mentioned or at least constitutes a highly plausible assumption (Givon, 1978). As in Glenberg et al.’s experiment, we manipulated the linguistic contexts in which